| Gary J. Nutt, Adam J. Griff, James E. Mankovich, and Jeffrey D. McWhirter, "Extensible Parallel Program Performance Visualization," International Workshop on Modeling, Analysis and Simulation of Computer and Telecommunication Systems (MASCOTS'95) |
....2. Related Work The goal of the ULTRA instrumentation is to provide accurate characterization of the intrinsic characteristics of large, parallel message passing programs that can be applied to other computer architectures, rather than to tune performance (e.g. Paradyn [Miller95] Paravision [Nutt95], and Pablo [Reed94] which focuses on identifying performance problems in code on existing computers. Thus, most of the performance tools focus on time based metrics because they try to identify how a program spends 4 10:04:16 October 1, 1999 its time. However, the performance bottlenecks on ....
Gary J. Nutt, Adam J. Griff, James E. Mankovich, and Jeffrey D. McWhirter, "Extensible Parallel Program Performance Visualization," International Workshop on Modeling, Analysis and Simulation of Computer and Telecommunication Systems (MASCOTS'95)
....has properly terminated. Any number of solutions to this problem are possible: The events can be written to separate files based on process id (this causes other subtle complications in processing and visualizing the events later) The event file can be post processed to remove logical inaccuracies[BS93, NMGM95]; A logical time tracing approach can be used[TC93, TSS94] Perhaps the easiest of these methods is to post process one trace file. Even if the logical time errors are corrected, this approach still has another problem. When the one file is read and visualized by threadview, an essentially serial ....
Gary J. Nutt, James E. Mankovich, Adam J. Griff, and Jeffrey D. McWhirter. Extensible parallel program performance visualization. In Proceedings of the MASCOTS '95 International Workshop on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, pages ??--??, Durham, NC, January 1995.
....is that thread synchronizations may create serialization bottlenecks which are often hard to detect. Removing serialization bottlenecks is referred to as performance tuning. Different tools for visualizing the behaviour of, and thus the bottlenecks in, parallel programs have been developed [1, 2, 3, 5, 6, 9, 10, 11, 12, 14, 16, 18]. The tuning process may benefit significantly from using such tools. Some performance visualization tools show the behaviour of one particular monitored multiprocessor execution of the parallel program [1, 2, 3, 5, 6, 9, 10] If we monitor the execution on a multiprocessor with four processors ....
....bottlenecks in, parallel programs have been developed [1, 2, 3, 5, 6, 9, 10, 11, 12, 14, 16, 18] The tuning process may benefit significantly from using such tools. Some performance visualization tools show the behaviour of one particular monitored multiprocessor execution of the parallel program [1, 2, 3, 5, 6, 9, 10]. If we monitor the execution on a multiprocessor with four processors such tools make it possible to detect bottlenecks which are present when using four processors. The problem with this approach is the lack of support for detecting bottlenecks which appear when using another number of ....
G. J. Nutt, A. J. Griff, J. E. Mankovich, and J. D. McWhirter, "Extensible Parallel Program Performance Visualization," Proc. Mascots `95, 1995.
....possibly heterogeneous, tools that cooperate for carrying out one or more analyses of the same parallel program. Tools built by different developers are referred to as heterogeneous tools by Hao et al. 7] An integrated environment may support off line tool usage, such as TAU [2] and ParaVision [20]; homogeneous on line tool usage, such as Paradyn [19] or a combination of the two, such as SPI [1] VIZIR [7] and ParAide [24] Malony [18] presents a classification of measurement based tools comprising four classes: profilebased (sampling) trace based, prediction based, and automated ....
Nutt, Gary J. and Adam J. Griff, "Extensible Parallel Program Performance Visualization," Proc. of Int. Workshop on Modeling, Analysis and Simulation of Computer and Telecommunication Systems (MASCOTS `95), Durham, North Carolina, Jan. 1995.
....database approach we propose can be extended to provide support for object oriented parallel codes. Visualization. A great deal of effort has gone into performance visualization. There are a large number of tools on many platforms that provide visual displays of performance data [10] 11] 9] [27]. Similar to data scalability, there is the notion of visual scalability. Most visualizations work well for programs that run for a short time on a small number of processors. As these quantities increase, the visual display becomes less useful and more overwhelming to the user. Support for ....
G. J. Nutt, A. J. Griff, J. E. Mankovich, and J. D. McWhirter. Extensible parallel program performance visualization. In Proceedings of the International Workshop on Modeling Analysis and Simulation of Computer and Telecommunication Systems 3rd, pages pp. 205--211, Durham, NC, January 1995.
....software includes performance monitoring utility http: www.CS.Princeton.EDU shrimp TAU Parallel systems using pC Performance tuning and analysis http: www.cs. uoregon.edu paracomp tau ParaVision Parallel systems Integrated environment of performance evaluation, debugging, and visualization [15] ParAide Distributed memory parallel systems Integrated environment of performance evaluation, debugging, and visualization http: www.ssd.intel.com paragon.html Prism Distributed memory parallel systems Integrated environment of performance evaluation, debugging, and data visualization ....
GARY J. NUTT and ADAM J. GRIFF, "Extensible Parallel Program Performance Visualization," Proc. of MASCOTS `95, (Durham, North Carolina, Jan. 1995).
No context found.
Gary J. Nutt, Adam J. Griff, James E. Mankovich, & Jeffrey D. McWhirter. (1995). Extensible Parallel Program Performance Visualization. International Workshop on Modeling, Analysis and Simulation of Computer and Telecommunication Systems (MASCOTS'95). Available URL: http://www.cs.colorado.edu/~griff/CV/Mascots95.pdf
No context found.
G. J.Nutt, A. J.Griff, J. E.Mankovich, and J. D. McWhirter, Extensible parallel program performance visualization, Tech. Report CB 430, Department of Computer Science, University of Colorado, Boulder, CO.
No context found.
Nutt, Gary J. and Adam J. Griff, "Extensible Parallel Program Performance Visualization," Proc. of Int. Workshop on Modeling, Analysis and Simulation of Computer and Telecommunication Systems (MASCOTS `95), Durham, North Carolina, Jan. 1995.
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC